Subcribe and Access : 5200+ FREE Videos and 21+ Subjects Like CRT, SoftSkills, JAVA, Hadoop, Microsoft .NET, Testing Tools etc..

Batch Date: June 18th @8:00AM

Faculty: Mr. N. Vijay Sunder Sagar (20+ Yrs of Exp,..)

Duration : 2.5 Months

Venue :
DURGA SOFTWARE SOLUTIONS,
Flat No : 202, 2nd Floor,
HUDA Maitrivanam,
Ameerpet, Hyderabad - 500038

Ph.No: +91 - 8885252627, 9246212143, 80 96 96 96 96

Syllabus:

Full Stack Generative AI & Agentic AI
Build & Deploy Production Level AI Agents
with Hands On Projects

Module 1: Generative AI

• What is Generative AI?
• Generative AI Evolution
• Differentiating Generative AI from Discriminative AI
• Types of Generative AI
• Generative AI Core Concepts
• LLM Modelling Steps
• Transformer Models: BERT, GPT, T5
• Training Process of an LLM Model like ChatGPT
• The Generative AI development lifecycle
• Overview of Proprietary and Open Source LLMs
• Overview of Popular Generative AI Tools and Platforms
• Ethical considerations in Generative AI
• Bias in Generative AI outputs
• Safety and Responsible AI practices

Module 2: Prompt Engineering

• Introduction to Prompt Engineering
• Structure and Elements of Prompts
• Zero-shot Prompting
• One-shot Prompting
• Few-shot Prompting
• Instruction Tuning Basics
• Prompt Testing and Evaluation
• Prompt Pitfalls and Debugging
• Prompts for Different NLP Tasks (Q&A, Summarization, Classification)
• Understanding Model Behavior with Prompt Variation

Module 3: Advanced Prompting Techniques

• Chain-of-Thought (CoT) Prompting
• Tree-of-Thought (ToT) Prompting
• Self-Consistency Prompting
• Generated Knowledge Prompting
• Step-back Prompting
• Least-to-Most Prompting
• Adversarial Prompting & Prompt Injection
• Auto-prompting techniques
• Prompt testing and validation methodologies

Module 4: Working with LLM APIs

• LLM Landscape: OpenAI, Anthropic, Gemini, Mistral API, LLaMA
• Core Capabilities: Summarization, Q&A, Translation, Code Generation
• Efficient Use of Tokens and Context Window
• Calling Tools
• Functions With LLMs
• Deployment Considerations for Open-Source LLMs (Local, Cloud, Fine-Tuning)
• Rate Limits, Retries, Logging
• Understanding Cost, Latency, and Performance and Calculating via Code

Module 5: Building LLM Apps with LangChain &LlamaIndex

• LangChain Overview
• LlamaIndex Overview
• Building With LangChain: Chains, Agents, Tools, Memory
• Understanding LangChain Expression Language (LCEL)
• Working With LlamaIndex: Document Ingestion, Index Building, Querying
• Integrating LangChain and LlamaIndex: Common Patterns
• Using External APIs and Tools as Agents
• Enhancing Reliability: Caching, Retries, Observability
• Debugging and Troubleshooting LLM Applications

Module 6: Developing RAG Systems

• What is RAG and Why is it Important?
• Addressing LLM limitations with RAG
• The RAG Architecture: Retriever, Augmenter, Generator
• DocumentLoaders
• Embedding Models in RAG
• Customizing Prompts for RAG
• Advanced RAG Techniques: Re-ranking retrieved documents
• Query Transformations
• Hybrid Search
• Parent Document Retriever and Self-Querying Retriever
• Evaluating RAG Systems: Retrieval Metrics

Module 7: Vector Databases and Embeddings

• What are Text Embeddings?
• How LLMs and Embedding Models generate embeddings
• Semantic Similarity and Vector Space
• Introduction to Vector Databases
• Key features: Indexing, Metadata Filtering, CRUD operations
• ChromaDB: Local setup, Collections, Document and Embedding Storage
• Pinecone: Cloud-native, Indexes, Namespaces, and Metadata filtering
• Weaviate: Open-source, Vector-native, and Graph Capabilities
• Other Vector Databases: FAISS, Milvus, Qdrant
• Vector Indexing techniques
• Data Modeling in Vector Databases
• Updating and Deleting Vectors
• Choosing the Right Embedding Model
• Evaluation of Retrieval quality from Vector Databases

Module 8: Building End-to-End GenAI Applications

• Architecting LLM-Powered Applications
• Types of GenAI Apps: Chatbots, Copilots, Semantic Search / RAG Engines
• Design Patterns: In-Context Learning vs RAG vs Tool-Use Agents
• Stateless vs Stateful Agents
• Modular Components: Embeddings, VectorDB, LLM, UI
• Key Architectural Considerations: Latency, Cost, Privacy, Memory, Scalability
• Building GenAI APIs with FastAPI
• RESTful Endpoint Structure
• Async vs Sync, CORS, Rate Limiting, API Security
• Orchestration Tools: LangServe, Chainlit, Flowise
• Cloud Deployment: GCP
• Containerization and Environment Setup

Module 9: Evaluating GenAI Applications and Enterprise Use Cases

• Evaluation Metrics: Faithfulness, Factuality, RAGAs, BLEU, ROUGE, MRR
• Human and Automated Evaluation Loops
• Logging, Tracing, and Observability Tools: LangSmith, PromptLayer, Arize
• Prompt and Output Versioning
• Chain Tracing and Failure Monitoring
• Real-Time Feedback Collection
• GenAI Use Cases: Customer Support, Legal, Healthcare, Retail, Finance
• Contract Summarization
• Legal Q&A Bots
• Invoice Parsing with RAG
• Product Search Applications
• Domain Adaptation Strategies

Module 10: Multimodal LLMs

• Introduction to Multimodal LLMs (GPT-4V, LLaVA, Gemini)
• How multimodal models process different data types
• Use Cases: Image Captioning, Visual Q&A, Video Summarization
• Working with Vision-Language Models (VLMs): Image inputs, text outputs
• Image Loaders in LangChain/LlamaIndex
• Simple visual Q&A applications
• Audio Processing with LLMs: Speech-to-Text (ASR)
• Text-to-Speech (TTS) integration
• Video understanding with LLMs
• Challenges in Multimodal AI
• Ethical Considerations in Multimodal AI
• Agent Frameworks (AutoGPT, CrewAI, LangGraph, MetaGPT)
• ReAct and Plan-and-Act agent strategies
• Future Direction

Module 11: LLMOps and Evaluation

• Introduction to LLMOps: Managing the ML Lifecycle for Large Language Models
• Introduction to Model Finetuning: When Prompt Engineering Isn’t Enough
• Overview of Parameter-Efficient Finetuning (PEFT)
• LoRA (Low-Rank Adaptation): Concept and Architecture
• QLoRA: Quantized LoRA for Finetuning Large Models Efficiently
• Adapter Tuning: Modular and Lightweight Finetuning
• Comparing Finetuning Techniques: Full vs. LoRA vs. QLoRA vs. Adapters
• Selecting the Right Finetuning Strategy Based on Task and Resources
• Introduction to Hugging Face Transformers and PEFT Library
• Setting Up a Finetuning Environment with Google Colab
• Preparing Custom Datasets for Instruction Tuning and Task Adaptation
• Monitoring Training Metrics and Evaluating Fine-tuned Models
• Use Cases: Domain Adaptation, Instruction Tuning, Sentiment Customization

Module 12: Agentic AI

• Agentic AI Introduction
• AI Agents vs. Agentic AI
• Comparison: Agentic AI, Generative AI, and Traditional AI
• Agentic AI Building Blocks
• Autonomous Agents
• Human in the Loops Systems
• Single and Multi Agent AI Systems
• Agentic AI Frameworks Overview
• Ethical and Responsible AI
• Agentic AI Best Practices

Module 13: Agentic AI: Architectures and Design Patterns

• Agentic AI Introduction
• AI Agents vs. Agentic AI
• Comparison: Agentic AI, Generative AI, and Traditional AI
• Agentic AI Building Blocks
• Autonomous Agents
• Human in the Loops Systems
• Single and Multi Agent AI Systems
• Agentic AI Frameworks Overview
• Ethical and Responsible AI
• Agentic AI Best Practices

Module 14: Working with LangChain and LCEL Topics

• Components and Modules
• Data Ingestion and Document Loaders
• Text Splitting
• Embeddings
• Integration with Vector Databases
• Introduction to Langchain Expression Language (LCEL)
• Runnables
• Chains
• Building and Deploying with LCEL
• Deployment with Langserve

Module 15: Building AI Agents with LangGraph Topics

• Introduction to LangGraph
• State and Memory
• State Schema
• State Reducer
• Multiple Schemas
• Trim and Filter Messages
• Memory and External Memory
• UX and Human-in-the-Loop (HITL)
• Building Agent with LangGraph
• Long Term Memory
• Short vs. Long Term Memory
• Memory Schema
• Deployment

Module 16: Implementing Agentic RAG

• What is Agentic RAG?\
• Agentic RAG vs. Traditional RAG
• Agentic RAG Architecture and Components
• Understanding Adaptive RAG
• Variants of Agentic RAG
• Applications of Agentic RAG
• Agentic RAG with Llamaindex
• Agentic RAG with Cohere

Module 17: Developing AI Agents with Phidata

• Agents
• Models
• Tools
• Knowledge
• Chunking
• Vector DB
• Storage
• Embeddings
• Workflows
• Developing Agents with Phidata

Module 18: Multi Agent Systems with LangGraph CrewAI

• Multi Agent Systems
• Multi Agent Workflows
• Collaborative Multi Agents
• Multi Agent Designs
• Multi Agent Workflow with LangGraph
• CrewAI Introduction
• CrewAI Components
• Setting up CrewAI environment
• Building Agents with CrewAI

Module 19: Advanced Agent Development with Autogen

• Autogen Introduction
• Salient Features
• Roles and Conversations
• Chat Terminations
• Human-in-the-Loop
• Code Executor
• Tool Use
• Conversation Patterns
• Developing Autogen-powered Agents
• Deployment and Monitoring

Module 20: AI Agent Observability and AgentOPs

• AI Agent Observability and AgentOPs
• Langfuse Dashboard
• Tracing
• Evaluation
• Managing Prompts
• Experimentation
• AI Observability with Langsmith
• Setting up Langsmith
• Managing Workflows with Langsmith
• AgentOps Practical Implementation

Module 21: Building AI Agents with No/Low- Code Tools

• Introduction to No-Code/Low-Code AI
• Benefits and Challenges of No-Code AI Development
• Key Components of No-Code AI Platforms
• Building AI Workflows Without Coding
• Designing AI Agents with Drag-and-Drop Interfaces
• Integrating No-Code AI with Existing Systems
• Customizing and Fine-Tuning AI Solutions
• Optimizing Performance and Efficiency in No-Code AI
• Security and Compliance Considerations in No-Code AI
• Best Practices for Deploying No-Code AI Solutions
• Real-World Use Cases and Applications of No-Code AI
• calling and Future Trends in No-Code AI